Syst. Biol. 53(5):711–734, 2004 ISSN: 1063-5157 print ... · Syst. Biol. 53(5):711–734, 2004...

Syst. Biol. 53(5):711–734, 2004Copyright c© Society of Systematic BiologistsISSN: 1063-5157 print / 1076-836X onlineDOI: 10.1080/10635150490468710

The Origin and Radiation of Macaronesian Beetles Breeding in Euphorbia: The RelativeImportance of Multiple Data Partitions and Population Sampling

BJARTE H. JORDAL1,2 AND GODFREY M. HEWITT1

1School of Biological Sciences, University of East Anglia, NR47TJ Norwich, United Kingdom2(Current Address) Museum of Natural History and Archaeology, Section of Natural history, NTNU, NO-7491, Norway;

E-mail: [email protected]

Abstract.—Species-level phylogenies derived from many independent character sources and wide geographical samplingprovide a powerful tool in assessing the importance of various factors associated with cladogenesis. In this study, weexplore the relative importance of insular isolation and host plant switching in the diversification of a group of bark beetles(Curculionidae: Scolytinae) feeding and breeding in woody Euphorbia spurges. All species in the genus Aphanarthrum are eachassociated with only one species group of Euphorbia (succulents or one of three different arborescent groups), and the majorityof species are endemic to one or several of the Macaronesian Islands. Hence, putative mechanisms of speciation could beassessed by identifying pairs of sister species in a phylogenetic analysis. We used DNA sequences from two nuclear andtwo mitochondrial genes, and morphological characters, to reconstruct the genealogical relationships among 92 individualsof 25 species and subspecies of Aphanarthrum and related genera. A stable tree topology was highly dependent on multiplecharacter sources, but much less so on wide population sampling. However, multiple samples per species demonstratedone case of species paraphyly, as well as deep coalescence among three putative subspecies pairs. The phylogenetic analysesconsistently placed the arborescent breeding and West African—Lanzarote–distributed species A. armatum in the most basalposition in Aphanarthrum, rendering this genus paraphyletic with respect to Coleobothrus. Two major radiations followed, onepredominantly African lineage of succulent feeding species, and one island radiation associated with arborescent host plants.Sister comparisons showed that most recent divergences occurred in allopatry on closely related hosts, with subsequentexpansions obscuring more ancient events. Only 6 out of 24 cladogenetic events were associated with host switching,rendering geographical factors more important in recent diversification. [Biogeography; data combinability; Euphorbia; hostplant use; molecular systematics; Scolytinae; speciation; taxon sampling.]

Reconstruction of species-level phylogenies is a pow-erful tool in the study of macroevolutionary patternsof diversification, where the assessment of putative sis-ter species provides a quantitative measure of the fac-tors associated with cladogenetic events. These measurescan only be as accurate as the phylogenies themselves,however, and various methodological problems may re-duce the precision needed for such studies. The bestpractical approach to overcome these problems involvesreconstructions under various optimality criteria, sam-pling every species and multiple populations of the fo-cus group, and sampling characters from as many inde-pendent character sources as necessary to stabilize thephylogenetic hypothesis. Most studies on higher levelphylogenies so far have sacrificed details for scale by in-cluding only one (or zero) individual per species, usingonly a single or a few sources of phylogenetic characters(Barraclough and Nee, 2001). Our aim in this study isto establish a robust phylogenetic hypothesis to exploreimportant factors in the diversification of a group of barkbeetles associated with Euphorbia spurges, by samplingfive different character partitions and nearly all speciesand populations in the reconstruction of the phylogeny.During the process of exploring these data, we investi-gate the effect of partial sampling of characters and taxa,providing empirical results to complement recent the-oretical discussions on these issues (Cummings et al.,1995; Poe and Swofford, 1999; Rosenberg and Kumar,2001; Pollock et al., 2002).

Putative diversifying factors associated with cladoge-netic events may be of two main categories, geograph-ical isolation or ecological shifts (Barraclough and Nee,

2001). The relative strength of the comparative approachto assess speciation modes, however, depends logicallyon how accurately geographical and ecological differ-ences can be defined, and how stable these are over evo-lutionary time scales (Losos and Glor, 2003). Perhaps themost precise model systems to explore patterns of colo-nization and geographic speciation are groups of speciesendemic to different oceanic islands, where each islandin an archipelago has very similar barriers of open waterbetween them (Juan et al., 2000; Emerson, 2002; Gillespieand Roderick, 2002). Similar or even stronger barriers togene flow, in terms of ecological isolation, occur in or-ganisms that specialize on narrow resources for feedingor reproduction. Some of the best examples on ecolog-ical isolation involves plants that serve as the sole hostfor herbivores, especially when the herbivores completetheir entire life cycle within a single resource (Kelley andFarrell, 1998; Machado et al., 2000; Pellmyr and Leebens-Mack, 2000; Scheffer and Wiegmann, 2000; Despres et al.,2002).

Whereas many phylogenetic studies to date have fo-cused on either geographical or ecological factors in spe-ciation, rather few have tried to include both in a si-multaneous study (Joy and Conn, 2001; Jordan et al.,2003; Percy, 2003). This paper explores the potential in-fluence from both of these factors in the diversificationof crypturgine bark beetles breeding in arborescent Eu-phorbia spurges on the Macaronesian islands. Macarone-sia is made up of four archipelagos west of the Africancoast: the Azores (which do not host any crypturginebeetles), Cape Verde, and the Canary and Madeiranisland groups. The oldest islands in each archipelago

711

at Zoological institute R

AS on M

arch 20, 2015http://sysbio.oxfordjournals.org/

Dow

nloaded from

http://sysbio.oxfordjournals.org/

712 SYSTEMATIC BIOLOGY VOL. 53

(Fuerteventura, Porto Santo, and the eastern Cape Verdeislands) are known to be at least 18 My, with youngerislands gradually decreasing in age westward, with themost recent island only 1.1 My (El Hierro; see Hess et al.,2000, and Brown et al., 2001). Hence, these islands pro-vide opportunities for diversification over an extendedperiod of time, similar to the Hawaiian and Galapagosarchipelagos, which have played prominent roles as evo-lutionary model systems (e.g., Gillespie and Roderick,2002).

Arborescent growth forms have evolved several timesindependently in Euphorbia (Molero et al., 2002), andare particularly frequent on islands and coastal areas,as in many other oceanic plant groups (Bohle et al., 1996;Panero et al., 1999). Their prevalence on the Macarone-sian islands and in drier parts of the African mainlandhas provided an unique opportunity for niche shifts andradiations in bark beetles of the weevil subfamily Scolyti-nae (Jordal et al., 2003). The largest clade of beetles asso-ciated with Euphorbia are found in the tribe Crypturgini,where all species of Aphanarthrum (19) and Coleoboth-rus (4) breed exclusively in dead or moribund parts ofthese plants. The majority of species are found on oneor several of the Macaronesian islands (17) or westernAfrica (5), with the remaining species (5) reaching east-ern parts of Africa and India. Five species have alsobeen divided into two or three subspecies based on gen-italic differences and allopatric island distributions (Is-raelson, 1972), suggesting that speciation is currently anongoing process. The closest relatives to Aphanarthrumand Coleobothrus are most likely Cisurgus and Crypturguswhose species are morphologically very similar to eachother (Wood, 1986). However, all species of the mainlyHolarctic Crypturgus (10) breed in conifers only, whereasspecies of the Mediterranean Cisurgus (6) breed in vari-ous woody herbs, including two species breeding in suc-culent euphorbs in Morocco and the Canary Islands. Thelast two genera in Crypturgini are the monotypic Dolur-gus (breeding in North American pines) and Deropria(associated with Rubus in the Canary Islands), believedto be distantly related to the other four genera (Wood,1986).

Dead or moribund plant material serves simultane-ously as both the food resource and as breeding spacefor adults and larvae of these beetles, where they com-plete the entire life cycle as in all other Scolytinae barkbeetles. The level of host plant fidelity is high, and eachspecies show distinct preference for only one of the fourmain groups of Euphorbia which have proven suitable ashost plants (Table 1; Jordal, submitted). All host plantsproduce various forms of latex (Carter and Smith, 1988;Seigler, 1994) that are still present during the early stageof colonization, and are further distinguished by growthform and structural differences in tissue (Webster, 1994)as illustrated by the cactus-like succulents (e.g., E. ca-nariense, E. handiense), and the shrub-like groups consist-ing of E. lamarckii complex (includes the artropurpurea-complex), E. balsamifera, and E. longifolia. The four groupsare also deeply separated by molecular data (Moleroet al., 2002), further confirming the different nature of

potential host plants. The wide and largely overlappingdistributions among each of the three first and mostimportant host groups, together with the beetle’s dis-tinct preference for one of the groups, make cladogeneticevents associated with host switching less ambiguous inthis model system.

Current classification of Crypturgini, with reciprocalmonophyly of all genera, implies at least three origins ofthe Macaronesian species and two origins of Euphorbiaassociations. The aim of this study is to establish a newand well-supported phylogeny to allow a more exactestimation of the rate and direction of island-mainlandtransitions and host plant switching, and hence, to esti-mate the relative frequency of allopatric divergence ver-sus host shifts in diversification. The largely overlappingdistribution within several guilds of ecologically simi-lar species, and the lack of externally diagnostic charac-ters between most of these (Israelson, 1972), may suggestthat not all of the species are reciprocally monophyletic.Recent splits often reveal problems with paraphyleticspecies, or worse, lineage sorting of ancestral polymor-phism or even reticulate evolution under low frequencyhybridization. To be able to assess these problems, if theydo occur, we included multiple specimens per species,from several different islands, and characters from fivedifferent partitions. We used sequences from two geneseach from the nuclear and the mitochondrial genomes,and morphological characters, to avoid biased gene treerepresentation of the species tree, and explored the rela-tive contribution of each partition to the various depthsof the combined data phylogeny.

MATERIAL AND METHODS

Taxon Sampling

We collected and sequenced 92 specimens, includingall Macaronesian and African species of Aphanarthrumand Coleobothrus (Table 1), two species each of the out-groups Crypturgus and Cisurgus, and Dolurgus pumilus(Mannerheim). Preliminary analyses of 839 EF-1α cod-ing characters, from 12 crypturgine and 45 additionalspecies of Scolytinae and Platypodine, confirmed thepreviously suggested monophyly of Crypturgini (Wood,1986), and that Dolurgus, and Crypturgus plus Cisurgus,are nested sets of outgroups suitable for all further anal-yses of a monophyletic Aphanarthrum plus Coleobothrus.An identical outgroup topology also resulted from theanalysis of 316 aligned characters from the large riboso-mal subunit (28S) for 23 crypturgine taxa and 3 scolytineoutgroups.

Collected specimens were compared to the originaldescriptions or type material in the Natural History Mu-seum, London, and Zoologische Staatssammlung, Mu-nich. Based on these comparisons, three species weretreated as synonyms as follows: A. saturatum Peyer-imhoff and A. duongi Villiers = A. mairei Peyerimhoff,and A. mododi Paulian & Villiers = A. armatum Wollas-ton. Also Cisurgus resiniferae Peyerimhoff was treated asa synonym of Ci. occidentalis Payerimhoff. Three Apha-narthrum species described from India (Wood, 1988) were


AS on M


Dow

nloaded from


2004 JORDAL AND HEWITT—RADIATION OF MACARONESIAN BEETLES 713

TABLE 1. Species and populations included in the phylogenetic analyses. Samples not collected by the senior author are indicated by collectorsname (in parentheses).

Species Locality Host

Outgroups:Dolurgus pumilus USA: AK, Juneau 24.6.2002 (M. Schultz) Pinus sitchensisCrypturgus borealis USA: NH, Mt. Monadnock, 26.5.1998 Picea sp.Cr. hispidulus Norway: Nordland, Svenningdal, 26.7.2001 Picea sp.Cisurgus occidentalis Morocco: S Agadir, Massa, 16.4.2002 E. echinusCi. occidentalis Morocco: N Agadir, Cap Rhir, 17.4.2002 E. echinusCi. occidentalis Morocco: S Agadir, Bouzacarne, 20.4.2002 E. echinusCi. wollastoni CI: Tenerife, Punta Teno, 6.8.1998 (L. Kirkendall) E. canarienseCi. wollastoni CI: El Hierro, Timijiraque, 7.3.2002 E. canarienseCi. wollastoni CI: La Palma, Punta del Eden, 4.3.2002 E. canariense

Ingroup:Aphanarthrum aeoni CI: La Gomera, Alajero, 10.3.2002 E. lamarckiiA. aeoni CI: La Gomera, El Cercado, 10.3.2002 E. lamarckiiA. affine CI: Lanzarote, Mirador de Haria, 27.2.2002 E. regis-jubaeA. affine Morocco: N Agadir, Cap Rhir, 17.4.2002 E. regis-jubaeA. affine Morocco: S Agadir, Sidi Ifni, 20.4.2002 E. regis-jubaeA. affine CI: Gran Canaria, Galdar, 22.2.2002 E. regis-jubaeA. affine CI: Fuerteventura, Lajares, 26.2.2002 E. regis-jubaeA. armatum CI: Lanzarote, Playa Blanca, 27.2.2002 E. balsamiferaA. bicinctum (vestitum) CI: Tenerife, Tamaimo, 11.8.1999 (K. Harkestad) E. lamarckiiA. bicinctum (vestitum) CI: La Gomera, Alajero, 14.8.1999 (K. Harkestad) E. lamarckiiA. bicinctum (vestitum) CI: La Gomera, San Sebastian, 8.3.2002 E. lamarckiiA. bicinctum (obsitum) CI: Gran Canaria, Galdar, 22.2.2002 E. regis-jubaeA. bicinctum (bicinctum) Morocco: N Agadir, Cap Rhir, 17.4.2002 E. regis-jubaeA. bicinctum (bicinctum) CI: Lanzarote, Mirador de Haria, 27.2.2002 E. regis-jubaeA. bicinctum (bicinctum) CI: Fuerteventura, Lajares, 26.2.2002 E. regis-jubaeA. bicolor CI: El Hierro, La Dehesa, 6.3.2002 E. lamarckiiA. bicolor CI: La Palma, El Time, 1.3.2002 E. lamarckiiA. bicolor CI: La Gomera, Vallehermoso, 9.3.2002 E. lamarckiiA. bicolor CI: Tenerife, San Marcos, 15.2.2002 E. lamarckiiA. bicolor CI: Tenerife, Bco. Balbyo, 12.3.2002 E. lamarckiiA. bicolor Ma: Porto Santo, 7.9.2002 E. piscatoriaA. bicolor Ma: Madeira, Pico das Furnas, 8.9.2002 E. piscatoriaA. canariense (canariense) CI: Tenerife, Los Cristianos, 16.2.2002 E. canarienseA. canariense (canariense) CI: Fuerteventura, Cofete, 25.2.2002 E. canarienseA. canariense (canariense) CI: La Gomera, Bco. Rincon E. canarienseA. canariense (canariense) CI: El Hierro, Timijiraque, 7.3.2002 E. canarienseA. canariense (neglectum) CI: La Palma, Playa de Nogales, 28.2.2002 E. canarienseA. canariense (neglectum) CI: La Palma, Punta del Eden, 4.3.2002 E. canarienseA. canescens (canescens) CI: La Gomera, Airport, Playa Santiago, 10.3.2002 E. balsamiferaA. canescens (canescens) CI: La Gomera, Bco. Tapachuga, 11.3.2002 E. balsamiferaA. canescens (canescens) CI: La Gomera, Punta Llana, 11.3.2002 E. balsamiferaA. canescens (polyspiniger) CI: Tenerife, El Escobonal, 12.8.1999 (K. Harkestad) E. balsamiferaA. canescens (polyspiniger) CI: Tenerife, Playa de la Gaviota, 12.3.2002 E. balsamiferaA. canescens (polyspiniger) CI: Fuerteventura, La Matilla, 26.2.2002 E. balsamiferaA. euphorbiae Ma: Madeira, Chao de Laoros, 5.9.2002 E. longifoliaA. euphorbiae Ma: Madeira, Encumeada, 5.9.2002 E. longifoliaA. euphorbiae Ma: Madeira, 1998 FIT trapA. glabrum (glabrum) CI: El Hierro, Ermita Virgen, 6.3.2002 E. lamarckiiA. glabrum (glabrum) CI: La Gomera, Alojera, 9.3.2002 E. lamarckiiA. glabrum (glabrum) CI: Tenerife, Bco. Balbyo, 12.3.2002 E. lamarckiiA. glabrum (glabrum) CI: Tenerife, Los Corrizales, 13.3.2002 E. artropurpureaA. glabrum (nudum) CI: La Palma, Brena Baja, 2.3.2002 E. lamarckiiA. glabrum (nudum) CI: La Palma, El Faro, 2.4.2002 E. lamarckiiA. hesperidum CV: Sao Vicente, Monte Verde, 7.10.2002 E. tuckeyanaA. hesperidum CV: Santo Antao, Cova de Paul, 10.10.2002 E. tuckeyanaA. hesperidum CV: Santo Antao, Rabo Curto, 9.10.2002 E. tuckeyanaA. jubae (jubae) CI: Fuerteventura, Pajara, 24.2.2002 E. balsamiferaA. jubae (jubae) CI: Lanzarote, Playa Blanca, 27.2.2002 E. balsamiferaA. jubae (jubae) CI: Tenerife, Roque de las Bodegas, 14.2.2002 E. balsamiferaA. jubae (tuberculatum) CI: El Hierro, Faro de Orchilla, 6.3.2002 E. balsamiferaA. jubae (tuberculatum) CI: El Hierro, Charco Manso, 6.3.2002 E. balsamiferaA. mairei Morocco: Massa, S Agadir, 16.4.2002 E. echinusA. mairei Morocco: Cap Rhir, N Agadir, 17.4.2002 E. echinusA. mairei CI: Fuerteventura, Jandia, Gran Valle, 25.2.2002 E. handienseA. orientalis Uganda: Fort Portal, 1.7.1998 E. tekeA. orientalis Uganda: Queen Elisabeth NP, 25.6.1998 E. teke

(Continued on next page)


AS on M


Dow

nloaded from



TABLE 1. Continued

Species Locality Host

A. piscatorium Ma: Madeira, Pico das Furnas, 8.9.2002 E. piscatoriaA. piscatorium Ma: Porto Santo, 7.9.2002 E. piscatoriaA. piscatorium CI: Gran Canaria, 4 km S Galdar, 22.2.2002s E. regis-jubaeA. piscatorium CI: Tenerife, Bco. Balbyo, 12.3.2002 E. lamarckiiA. piscatorium CI: La Gomera, Vallehermoso, 9.3.2002 E. lamarckiiA. piscatorium CI: El Hierro, La Guancha, 6.3.2002 E. lamarckiiA. piscatorium CI: La Palma, Mazo, 4.3.2002 E. lamarckiiA. pygmeum CI: Gran Canaria, 3 km S Tasarte, 20.2.2002 E. canarienseA. pygmeum CI: Tenerife, Los Cristianos, 16.2.2002 E. canarienseA. pygmeum CI: La Gomera, Bco. Rincon, 8.3.2002 E. canarienseA. pygmeum CI: La Palma, Playa de Nogales, 28.2.2002 E. canarienseA. pygmeum CI: El Hierro, Timijiraque, 7.3.2002 E. canarienseA. subglabrum CI: La Palma, El Time, 1.3.2002 E. lamarckiiA. subglabrum CI: La Palma, Punta del Eden, 4.3.2002 E. lamarckiiA. subglabrum CI: La Palma, Brena Baja, 2.3.2002 E. lamarckiiA. wollastoni CI: La Gomera, Bco. Rincon, 8.3.2002 E. canarienseA. wollastoni CI: La Gomera, El Cercado, 10.3.2002 E. lamarckiiA. wollastoni CI: La Gomera, Playa de la Caleta, 9.3.2002 E. lamarckiiA. wollastoni CI: La Gomera, Valle Gran Rey, 10.3.2002 E. lamarckiiColeobothrus allaudi Morocco: Cap Rhir, N Agadir, 17.4.2002 E. echinusCo. allaudi Morocco: Bouzacarne, 20.4.2002 E. echinusCo. germeauxi Uganda: Fort Portal, 1.7.1998 E. tekeCo. germeauxi Uganda: Queen Elisabeth NP, 25.6.1998 E. tekeCo. luridus CI: Gran Canaria, 3 km S Tasarte, 20.2.2002 E. canarienseCo. luridus CI: Tenerife, Punta Teno, 6.8.1998 (L. Kirkendall) E. canarienseCo. luridus CI: La Gomera, Bco. Rincon, 8.3.2002 E. canarienseCo. luridus CI: La Palma, El Time, 3.3.2002 E. canariense

not available for DNA extraction, but we note that thesespecies are very similar to the two Ugandan species in-cluded. Multiple specimens per species were includedfor all species possible, with coverage of nearly all is-lands within their known range. Current studies on thepopulation genetics of the Canary Island species con-firmed that our samples were unbiased with respect tothe populations included here.

Character Sampling

Morphological and behavioral characters.—Twenty-eightmorphological characters were coded and included inthe parsimony analyses (Appendices 1, 2). We also in-cluded Deropria elongata (Eggers) in this matrix to test themonophyly of all other crypturgine genera. Host plantuse and geographical distribution were based on litera-ture records (Israelson, 1972, 1976, 1979) and correctedand augmented after intensive sampling throughout theMacaronesian islands (Jordal, in preparation). Previousrecords from multiple Euphorbia groups were probablybased on rare stray specimens on the less preferredhost, as documented during recent sampling. Geograph-ical distributions of the beetles were largely congru-ent with literature records, with only minor correctionsneeded.

DNA amplification and sequencing.—We sequenced por-tions of two mitochondrial and two nuclear genes, usingmostly the same primer for sequencing as for PCR am-plification. Mitochondrial cytochrome oxidase I was am-plified with the primers s1718 5′-GGA GGA TTT GGAAAT TGA TTA GTT CC-3′ (forward), a2237 5′-CCG AATGCT TCT TTT TTA CCT CTT TCT TG-3′ (reverse), or

a2411 5′-GCT AAT CAT CTA AAA ACT TTA ATT CCWGTW G-3′ (reverse) (Normark et al., 1999). PCR reactionswere performed in a 25-µL volume containing 0.2 µM ofeach primer, 0.25 mM of each dNTP, 0.5 unit of BiolineBioTaq DNA polymerase, 1× buffer with MgCl2 to a finalconcentration of 1.5 mM. Typical PCR cycles consisted of90 s initial denaturing at 94◦C, followed by 38 cycles of94◦C for 30 s, 46◦C annealing for 60 s, and 72◦C extensionfor 60 s, followed by a final extension for 7 min.

The mitochondrial large ribosomal subunit, 16S, wasamplified under the same PCR conditions, using theprimers 16S 5′-TTT AAT CCA ACA TCG AGG-3′ (for-ward) and 16S 5′-CGC CTG TTT AAC AAA AAC AT-3′(reverse) (Cognato and Vogler, 2001).

The nuclear gene encoding Elongation Factor 1α wasamplified with the primers efs149 5′-ATC GAG AAGTTC GAG AAG GAG GCY CAR GAA ATG GG-3′ (for-ward) and efa1043 5′-GTA TAT CCA TTG GAA ATT TGACCN GGR TGR TT-3′ (reverse) (Normark et al., 1999). Weused a touchdown profile consisting of 43 cycles whereall cycles had 72◦C extension for 40 s and 94◦C denatur-ing for 30 s, with the first 16 cycles having decreasingannealing temperature from 60◦C to 47◦C by 2◦C (last1◦C) every second cycle, with the final 27 cycles at 46◦Cfor 60 s.

The nuclear encoded Enolase gene (2-phospho-D-glycerate hydrolase) (Tracy and Hedges, 2000; Farrellet al., 2001) was amplified with nested PCRs using ini-tially the primers ens57c 5′-GAC TCC CGT GGN AACCCC ACN GTG GAG GT-3′ (forward) and ena886 5′-CCA GTC RTC YTG RTC RAA XGG-3′ (reverse), withthe latter replaced by ena780 5′-TCT TGA AGT CCAAAT CGT A-3′ (reverse, sequencing primer) in the nested


AS on M


Dow

nloaded from



PCR. Initial reactions were performed under similarconditions as for the mtDNA genes, but with 2.5 mMMgCl2 concentration and 44◦C annealing temperature.The nested PCR reactions used 1.5 mM MgCl2 concen-tration, with 1.0 µg bovine serum albumin added. Cyclesconsisted of 90 s initial denaturing at 94◦C, followed by40 cycles where all cycles had 72◦C extension for 60 s and94◦C denaturing for 30 s, with the first 16 cycles havingincreasing annealing temperature from 40◦C to 55◦C by2◦C (last 1◦C) every second cycle, with the final 12 cyclesat 55◦C for 60 s. Purified PCR products were sequencedusing a standard protocol for big-dye version 3 (PerkinElmer).

Sequences were assembled and edited with the Laser-gene software (DNASTAR, Inc.). There are multiplecopies known for EF-1α and Enolase and we used thesingle-intron copy (C1) of EF-1α and intron free copy ofEnolase (Eno1) in this study (Farrell et al., 2001; Jordal,2002). Sequences can be downloaded from GenBankunder the accession numbers AY500900 to AY500991(EF-1α), AY514121 to AY514211 (Enolase), AY514212 toAY514303 (16S), AY514904 to AY514995 (COI).

Phylogenetic Analyses

Alignments.—For Enolase and the EF-1α coding region,alignments were unambiguous due to the lack of anyinsertions or deletions in the coding region. Three COIsequences had indels (in multiples of 3 bp), but none ofthese putative pseudogenes had stop codons and werecautiously included; these sequences grouped togetherwith conspecific indel free sequences (A. aeoni El Cer-cado; Co. allaudi, Morocco). Sequences of 16S and theEF-1α intron were aligned in ClustalX (Thompson et al.,1997) under a wide range of parameters to assess align-ment stability (gap cost 2, 4, 6, 8, 12 16; gap exten-sion equal or half the cost of gap opening; transitionsweighted equally or half the cost of transversions). Pre-liminary alignments were chosen based on their topolog-ical congruence with the topology resulting from the eli-sion matrices which combines all alignments from eachpartition into one matrix, providing a gradual downweighting of ambiguous sites (Wheeler et al., 1995; Lee,2001). Final alignments were produced under gap open-ing to extension cost of 6:3, with transitions half the costof transversions, resulting in 68 ambiguous sites for 16S,and 54 ambiguous sites in the much more variable EF-1αintron. These alignments were also the most congruentwith the unambiguously aligned portions of the EF-1αexon and 16S, respectively. A secondary structure modelfor insects (Buckley et al., 2000) was used to assist in edit-ing the final 16S alignment. All indel ambiguous regionswere located in the terminal loops of helices 68 (9 bp), 75(48 bp, including all but 5 bp of the helix), 81′ (8 bp), and84′ (4 bp). Various alignments (or inclusion/exclusion ofambiguous sites) had no influence in the analyses of com-bined data. Neither did gaps as a fifth character changethe combined data topology. Hence we scored gaps asmissing characters in the parsimony analyses, to makecomparisons to likelihood analyses more realistic.

Sequence pruning.—To assess the relative importanceof sampling multiple sequences per species, and to in-crease the speed in likelihood analyses, we pruned onesequence from each of the species which had an excess ofsequences before each heuristic parsimony search. Thephylogenetic depth and number of node switches wasnoted for each partition, genome, and the combined data,on their respective strict consensus trees.

Tree searching.—Paup∗ 4.0 (Swofford, 1999) was usedin most phylogenetic analyses (the complete datamatrix and Figure 1 can be downloaded from Tree-Base, accession number SN1709, http://www.treebase.org/treebase/). Under the maximum-parsimony (MP)criterion, we performed 500 random addition replicatesof heuristic searches for each of the data partitions andthe combined matrix. Bootstrap support for individualnodes was assessed by 200 bootstrap replicates (Felsen-stein, 1985) of 10 random additions per replicate, withuninformative characters excluded. Bremer support wascalculated as the number of additional steps necessaryto collapse a node (Bremer, 1994), using TreeRot v.2(Sorenson, 1999). As DeBry (2001) recently pointedout, the significance of this metric is dependent onthe number of informative characters assigned to eachnode. We have used minimum branch lengths as aproxy estimate of informative characters, and appliedDeBry’s Table 3 to assess approximate significance ofeach Bremer support value.

Maximum-likelihood (ML) analyses were performedon the pruned data matrix only (54 terminals), with pa-rameters estimated through the application of Model-test 3.06 (Posada and Crandall, 1998). This applicationtests a nested set of substitution models using either thelog-likelihood ratio test (LRT) or the Akaike informationcriterion (AIC). When the two statistics disagreed, we ap-plied the AIC criterion to select the best model for eachpartition, or for all data combined. Ten random addi-tions were used in each of the final likelihood searches.ML branch lengths were also optimised with the molecu-lar clock enforced, and likelihood scores were comparedto those obtained with no constraints using the LRT.

MrBayes 3.1 (Ronquist and Huelsenbeck, 2003) wasused in the Bayesian analyses of the combined data, witha GTR +�+ model of substitution (selected by Model-test) fitted to each molecular partition, and a rate variablemodel fitted to the morphological data. Parameters wereestimated on each of the genes and morphology (5 parti-tions), or to each coding position, EF-1α intron, 16S, andmorphology (12 partitions). Initial searches of 500,000generations were used to assess convergence in likeli-hood values obtained from four running Markov chains,sampling trees every 500 generations. Trees from the first200,000 generations were discarded in the final analysisof 1,000,000 generations, as a ‘burn-in’ before stability inlikelihood values was achieved.

Incongruence.—The incongruence length differencetest (ILD) (Farris et al., 1995) has been frequently appliedto assess combinability of data partitions. However,both empirical (Yoder et al., 2001) and simulated data(Barker and Lutzoni, 2002, and references therein) have


AS on M


Dow

nloaded from



FIGURE 1. Strict consensus of 48 most parsimonious trees of length 5014 steps (CI = 0.35, RI = 0.82) resulting from the unweighted analysisof all data. Bootstrap support values are shown above the nodes, with minimum branch length and Bremer support values below (ml:BS,‘significant’ support shown in bold). Capital letters in bold indicate nodes discussed in the text. Geographic origin of each terminal is indicatedby capitals: SV, Sao Vicente; SA, Santo Antao (Cape Verde); Mad, Madeira; PS, Porto Santo (Madeira); EH, El Hierro; LP, La Palma; LG, La Gomera;T, Tenerife; GC, Gran Canaria; F, Fuerteventura; L, Lanzarote (Canary Islands); Mor, Morocco; Ug, Uganda, US, United States of America; N,Norway. The line drawing illustrates Aphanarthrum armatum.


AS on M


Dow

nloaded from



demonstrated the potential failure of this test to allowcombination of data when they should be. Potential neg-ative effects from combining statistically incongruentgene partitions can now be tested in a Bayesian frame-work, by analyzing all data under a uniform model ofsubstitution versus a mixed model where each partitionfits their own model. In a parsimony framework,incongruence can further be explored by accumulatingdata from an increasing number of partitions, aimingtowards a predictive hierarchy; that is, it remains stablewith the addition of more data (Nixon and Carpenter,1996). Incongruence may then be quantified by the levelof node support in conflicting nodes across differentpartitions. A node-specific approach to detect conflictingtaxa and partitions simultaneously was implementedthrough the calculation of partitioned Bremer support(Baker and DeSalle, 1997; Baker et al., 1998), usingTreeRot 2.0 (Sorenson, 1999). We estimated the relativecontribution from each data partition to the combinedanalysis, using default settings with 20 random additionsequences for each node in each of the partitions.

The overall support from different data partitionsmay also be asymmetric, such that one partition per-form significantly worse on the topologies resulting fromthe other partitions, but not vice versa. We applied theWilcoxon sign rank test (parsimony) in Paup∗ to test thisasymmetry, with the critical value Bonferroni correctedfor 49 consecutive comparisons among partitions. Forcomparison, we also applied the Shimodaira-Hasegawatest (Shimodaira and Hasegawa, 1999) on the molecu-lar partitions under the likelihood criterion, using 1000bootstrap pseudoreplicates.

Lineages that were topologically unstable across dif-ferent analyses were also examined for potential devia-tion in their base frequencies. We used the base hetero-geneity test as implemented in PAUP∗, and performedtests on all sites, and on informative sites only, to de-tect confounding opposite bias in invariant sites (Buckleyet al, 2001). Log-determinant distance analysis (Lockhartet al., 1994) is the only tree reconstruction method thatis specifically designed to avoid attraction of unrelatedlineages with similar base composition. We used thismethod to explore the possibility that biased base fre-quencies had any effect on conflicting or unstable nodes.

Character evolution.—MacClade 3.04 (Maddison andMaddison, 1992) was used for editing matrices, calculat-

TABLE 2. Tree statistics from the parsimony analyses of the various data partitions (54 taxa). Informative characters for the ingroup only(Aphanarthrum and Coleobothrus) are given in parenthesis. Incongruence refers to the number of node shifts compared to the combined data treetopology.

Partition Characters Informative Treelength CI RI No. of trees Incongruence

All data 2588 917 (860) 4116 0.37 0.76 2 —Molecular 2560 893 (840) 4050 0.37 0.75 1 0MtDNA 946 367 (353) 1969 0.33 0.72 3 3NucDNA 1614 526 (487) 2057 0.41 0.79 1 2EF-1α 927 272 (252) 858 0.48 0.82 16 4Enolase 687 254 (235) 1182 0.36 0.77 9 316S 454 154 (142) 618 0.43 0.81 8 1COI 492 213 (211) 1320 0.28 0.67 1 14Morphology 28 24 (20) 55 0.76 0.95 294 7

ing substitution frequencies, and tracing character trans-formations over the most parsimonious or likely treetopologies from the combined data analyses. We ap-plied the “chart state changes” option to count the min-imum and maximum transformations in host plant useand island-mainland distribution under ACCTRAN andDELTRAN optimisations. Although accelerated charac-ter transformations may be preferred due to conformingmore closely to primary homology statements (by im-plying reversals rather than parallel evolution of char-acters), there is no particular reason to expect one orthe other for labile island-mainland transitions or hostswitching.

Ancestral character states were also reconstructed us-ing maximum likelihood, incorporated in the computerprogram Discrete (Pagel, 1994, 1999). The likelihood ap-proach has the advantage (over parsimony methods)of estimating the probability of each ancestral charac-ter state, for each ancestor at a time, taking characterfrequencies and branch lengths (but not prior node esti-mates) into account.

RESULTS

Combined Analysis of All Data

The parsimony analysis of all 2688 characters (2660aligned base pairs, 28 morphological characters; Table 2),resulted in two trees differing only in the placement ofthe two El Hierro specimens of A. jubae (Fig. 1). Apha-narthrum plus Coleobothrus were monophyletic in thisanalysis, with the latter nested within Aphanarthrum. Thepredefined hypothesis on reciprocal monophyly of Apha-narthrum and Coleobothrus added 55 steps to the treelength, and was firmly rejected by the Kishino-Hasegawaand Wilcoxon tests (t = 4.01, z = −3.99, P < 0.001).

All but one of the species with multiple individu-als sampled were monophyletic, and most sister specieswere clearly defined. Aphanarthrum glabrum was para-phyletic with respect to A. subglabrum which formed thesistergroup to the La Palma subspecies of A. glabrum(nudum). Among the higher level relationships, only thepositions of A. jubae, and the internal arrangement of twoColeobothrus species with respect to A. mairei, were uncer-tain as shown by minimum Bremer and bootstrap sup-ports. Aphanarthrum armatum was the basal lineage, withall species associated with succulent Euphorbia in one


AS on M


Dow

nloaded from



well supported clade (C), and all Macaronesian speciesbreeding in arborescent euphorbs making up a highlysupported sistergroup (clade D). All clades that receivedmaximum bootstrap support had also significantly highBremer support, with only two significantly supportednodes having bootstrap values below 80%. There was agenerally high correlation between the two node supportindices for the combined data.

Outgroup manipulation.—Because the Dolurgus pumilussequences were relatively more diverged from the in-group than the Crypturgus and Cisurgus sequences, weexcluded this taxon in the remaining analyses. We alsoexcluded Cr. borealis because this taxon had the en-tire Enolase partition missing. The exclusion of thesetaxa did not change the topology resulting from thecombined parsimony analysis (or any of the molecu-lar, nucDNA, or EF-1α topologies). Support for the mostbasal nodes (A to G) increased after exclusion of D.pumilus and Cr. borealis, further suggesting that the analy-ses should be improved by using more narrowly definedoutgroups.

Sequence Pruning

Rather few contradicting changes occurred with in-creasingly pruned data matrices in each of the gene par-titions (Table 3); all data combined, nucDNA, and the16S data did not produce any contradictory nodes withincreased pruning of sequences. Most changes occurredin the two genes with highest substitution rates, Enolaseand COI (cf. Fig. 4), but these changes were unstable andsometimes reversed to the original topology in the nextstep of pruning. Overall, there were only marginal ef-fects of pruned matrices in each gene partition, downto three sequences per species, and all bootstrap sup-ported nodes (>50%) were recovered when only one ortwo sequences per species or subspecies were retained.Hence, we used two sequences per species or subspeciesin all further comparative analyses of the various genepartitions.

Separate Analyses—Parsimony

Morphology.—The analysis of morphological charac-ters resulted in 294 equally parsimonious trees of length55 steps (Table 2). The inclusion of all outgroups andDeropria elongata resulted in an identical topology forthe ingroup (Aphanarthrum plus Coleobothrus), and con-firmed the distant relationship of D. elongata in Cryp-turgini (Fig. 2) as previously suggested (Wood, 1986).

TABLE 3. Number of shifted interspecific nodes (total/deep only) after deletion of sequences. Only those taxa with excess sequences hadsequences deleted, one per species (or subspecies) before each search. The first column shows the number of sequences retained per species.Deep nodes were defined as clades A to D (see Fig. 1).

Sequences All data nDNA mtDNA EF-1α Enolase 16S COI Total

5 0/0 0/0 0/0 0/0 0/0 0/0 1/1 1/14 0/0 0/0 1/0 0/0 0/0 0/0 3/2 4/23 0/0 0/0 0/0 0/0 2/0 0/0 2/1 4/12 0/0 0/0 1/0 0/0 4/2 0/0 0/0 5/21 0/0 0/0 3/1 2/0 1/0 0/0 3/1 10/2

FIGURE 2. Strict consensus of 432 MP trees of length 73 steps each(CI = 0.71, RI = 0.87), resulting from the parsimony analysis of morpho-logical characters. Deropria elongata was included in this analysis to testthe monophyly of the ingroup. Bootstrap support values are shownabove the nodes. Black bars marked by numbers in squares indicatecharacters with synapomorphic character states (except autapomor-phies also shown in Deropria and Dolurgus, see Appendices 1 and 2).The line drawing illustrates Aphanarthrum armatum.

Our morphological data did not enable us to distinguishCrypturgus andCisurgus, but grouped Aphanarthrum andColeobothrus with high support (including two uniquelyderived characters), with the latter genus nested withinAphanarthrum. Topological congruence with the com-bined data tree was further exemplified by the basalplacement of A. armatum, the close relationship be-tween A. mairei, A. orientalis, and Coleobothrus, and the


AS on M


Dow

nloaded from



well-supported clade of Macaronesian shrub feeders(clade D). The lack of resolution in the latter clade wasapparently due to the low number of informative char-acters at this topological level.

Molecules.—The four molecular partitions contributedfrom 154 (16S) to 272 (EF-1α) informative characters(Table 2), and produced partly resolved topologies foreach partition, except for the fully resolved COI data(Fig. 3). Although several discrepancies appeared amongthe four consensus trees (Table 2), only three topologi-cal differences were supported by a bootstrap supportabove 70%, and only barely so (EF-1α: A. orientalis, A.bicinctum, and A. canescens. COI: A. bicolor). Major differ-ences of low or no support were mostly observed in theCOI topology where the basal structure was strikinglyabsent when compared to the combined data tree. Othernoteworthy differences were seen in Enolase, which dif-fered by the changed position of the otherwise basal A.armatum and the more basal position of A. orientalis, andin EF-1α, which differed in their basal position of A. eu-phorbiae plus A. affine in clade D.

The Enolase and COI genes had also significant biasesin base composition across taxa, caused by nonstation-arity in third positions (Table 4). EF-1α had a similarlysignificant bias in third position, but all parsimony in-formative sites combined were not significantly biasedfor this gene. For EF-1α and Enolase, the outgroups andA. armatum had generally intermediate base frequen-cies, and all succulent feeding taxa, and all remainingshrub feeding taxa, had divergent base frequencies thatwere more similar within each of these clades. The pat-tern was less clear in COI, and varied much more be-tween close relatives. We further noticed that A. orien-talis had frequencies more similar to the Coleobothrusclade than to the A. canariense clade for the two nucleargenes. However, a LogDet distance analysis of the nu-clear DNA partitions did not remove A. orientalis fromthe Coleobothrus clade. One exception to the within-cladecorrelation in base composition was found for the twobasal shrub feeding taxa in the EF-1α analysis, A. affineand A. euphorbiae, which had EF-1α base compositionsimilar to the succulent taxa and the outgroups. LogDetanalysis of these data placed A. bicinctum as the mostbasal taxon in this clade, consistent with most otheranalyses.

We applied the consistency index (CI) and the reten-tion index (RI) as measures of homoplasy (Table 2). Thesetwo correlated well with each other, and also reflected thedifferent distributions of substitution frequencies in thefour genes (Fig. 4) such that those genes with the fewestsubstitutions per character also had the highest consis-tency and synapomorphy (RI) indices.

We also used the partitioned Bremer support index(PBS) to measure the relative support of each partition tothe various depths of the phylogeny (Baker and DeSalle,1997). All four genes contributed positively to the com-bined data tree, although COI contributed consistentlyless at higher level nodes (Fig. 5, Table 5). This trend wasparticularly clear when PBS was corrected for minimumtree lengths for each partition.

Separate Analyses—Maximum Likelihood

All models selected by Modeltest were derivativesof a general time reversible model, where all assumedat least two different transition rates and two differ-ent transversion rates (Table 6). Major differences werefound between the nuclear and mitochondrial genes inthe guanine base frequencies and guanine-thymine sub-stitution frequencies. Also, the shape of the gamma-distributed substitution rates was higher in the Eno-lase partition, indicating a more uniform distribution ofcharacter changes for variable sites (cf. Fig. 4). Whenall sites were considered (including invariable sites), allgene partitions had nearly identical shape parameters(Table 4).

Overall, the ML analyses confirmed our results fromthe parsimony analyses, with a few noteworthy excep-tions (Fig. 6). In both of the Enolase and 16S partitions,A. canescens was moved to the basal position in clade D,and the position of A. orientalis was changed. Comparedto the combined data topology, several improvementsoccurred in the COI partition, with monophyly of A. eu-phorbiae and A. affine, and of all Coleobothrus species (in-cluding A. mairei). Aphanarthrum armatum moved closerto the base of the COI tree, and moved to the ‘correct’basal position in the Enolase tree. However, apparentlyworse results also occurred, with paraphyly of A. c. ca-nariense (EF-1α), A. wollastoni (Enolase), and the otherwisemonophyletic A. wollastoni plus A. aeoni (COI). Overall,the number of topological differences compared to thelikelihood analysis of all data combined was higher thanfor the equivalent parsimony analyses, with 6, 4, 6, and12 incongruent nodes found in the EF-1α, Enolase, 16S,and COI partitions, respectively.

Dynamics of Data Combination

Parsimony.—Combination of characters from the samegenome resulted in two consensus trees which only dif-fered from the combined data tree by two (nucDNA) orthree (mtDNA) topological positions (Fig. 7). All nodessupported by more than 70% bootstrap support in eachof the two genomic partitions were fully congruent witheach other and with the combined data tree. Bootstrapsupport in the nucDNA partition incorporated comple-mentary support from each of the Enolase and EF-1α par-titions, whereas some support in the mtDNA tree waslost, particularly for clades B and C, compared to theseparate 16S analysis. The combination of all molecularpartitions (Fig. 8) resulted in an overall increased boot-strap support for the most basal nodes (A to C), as well assome of the more derived clades (E to G). Adding mor-phological characters did not influence tree topology, ornode support, at all.

None of the separate or combined data partitionswere incongruent when forced upon the combined datatree (Table 7). Only the COI data were incongruent onthe nucDNA topology, whereas all nucDNA partitionswere incongruent upon the mtDNA topology. EF-1αhad the most congruent topology with only two par-titions (mtDNA, COI) being significantly incongruent.


AS on M


Dow

nloaded from



FIGURE 3. Strict consensus trees resulting from the parsimony analyses of each gene partition. Bootstrap support values are shown abovethe nodes, transitions in habitat selection under ACCTRAN optimisation is marked below nodes (A, arborescent; S, succulent). See Table 2 forfurther details. Selected taxa that varied in their position across analyses of different partitions are indicated by different symbols and fonts.


AS on M


Dow

nloaded from


TAB

LE

4.Su

mm

ary

ofse

quen

cest

atis

tics

for

the

vari

ous

mol

ecul

arch

arac

ter

part

itio

ns.T

ests

ond

evia

tion

from

base

stat

iona

rity

wer

epe

rfor

med

onal

lsit

es,o

rin

form

ativ

esi

tes

only

,an

dB

onfe

rron

icor

rect

edfo

r11

cons

ecut

ive

and

nono

verl

appi

ngd

ata

part

itio

ns(c

riti

calP

=0.

0045

,und

erlin

ed).

The

tran

siti

onto

tran

sver

sion

rate

(Ti:T

v)an

dth

esh

ape

ofth

e�

para

met

er(α

)wer

ees

tim

ated

und

era

HK

Y+�

mod

elof

subs

titu

tion

,wit

hin

vari

ants

ites

arti

ficia

llyse

tto

zero

for

com

pari

sons

betw

een

part

itio

ns.

EF-

1αE

nola

se16

SC

OI

1st

2nd

3rd

Intr

onA

ll1s

t2n

d3r

dA

llN

oga

psA

ll1s

t2n

d3r

dA

ll

Num

ber

posi

tion

s28

028

028

087

927

229

229

229

687

385

454

164

164

164

492

Var

iabl

epo

siti

ons

2813

183

7630

051

2421

228

710

916

441

1516

121

7In

form

ativ

epo

siti

ons

2212

169

6927

236

1420

425

410

015

440

1216

121

3B

ase

hom

ogen

eity

(P)

1.00

1.00

0.70

0.11

1.00

1.00

1.00

0.00

10.

001

1.00

1.00

1.00

1.00

0.00

00.

99π

A0.

290.

300.

260.

330.

290.

320.

320.

300.

320.

350.

360.

280.

170.

450.

30π

C0.

170.

240.

260.

120.

220.

110.

230.

240.

200.

200.

190.

240.

250.

190.

23π

G0.

390.

160.

120.

130.

220.

410.

150.

170.

240.

110.

100.

270.

160.

070.

17π

T0.

150.

290.

360.

420.

270.

160.

290.

290.

240.

340.

340.

210.

410.

290.

30B

ase

hom

ogen

eity

(inf

.onl

y)0.

991.

000.

001

0.03

0.60

0.89

1.00

0.00

00.

000

0.99

0.99

1.00

1.00

0.00

00.

000

πA

0.09

0.28

0.19

0.35

0.22

0.24

0.30

0.30

0.29

0.34

0.37

0.11

0.06

0.44

0.35

πC

0.38

0.38

0.32

0.13

0.29

0.18

0.43

0.25

0.25

0.17

0.16

0.43

0.36

0.19

0.25

πG

0.23

0.26

0.09

0.10

0.11

0.18

0.13

0.15

0.15

0.09

0.07

0.21

0.17

0.08

0.11

πT

0.30

0.08

0.40

0.42

0.38

0.40

0.14

0.30

0.31

0.40

0.40

0.24

0.41

0.29

0.29

Ti:T

vra

tio

2.66

1.02

4.83

0.85

2.42

3.48

0.55

2.59

2.44

1.75

1.70

17.1

41.

224.

972.

87α

shap

e0.

010.

010.

732.

240.

180.

150.

011.

760.

210.

150.

170.

110.

011.

440.

25

721


AS on M


Dow

nloaded from



FIGURE 4. Frequency distributions of character changes in the four molecular partitions. Invariable characters are not shown on the graphs.In 16S we distinguished between sites in alignment ambiguous and unambiguous regions.

Conversely, COI was the most incongruent topology,only congruent with the mtDNA data. The two mito-chondrial partitions were also less congruent than thetwo nuclear partitions as measured by the ILD test (COIversus 16S, P = 0.02; EF-1α versus Enolase, P = 0.08;mtDNA versus nucDNA, P = 0.001).

Maximum Likelihood.—The combination of all molec-ular data resulted in one tree which only differed intwo node positions compared to the parsimony analy-sis of combined data; A. orientalis was the sister groupto Coleobothrus, with Co. allaudi as the sister species ofA. mairei (Fig. 8). These nodes had very short internalbranches in the likelihood analyses, a possible explana-tion for the unstable positions of these taxa.

Shimodaira-Hasegawa tests of all molecular partitions(as listed in Table 7) produced quite different resultsfrom the Wilcoxon sign rank tests, with six new com-parisons significantly incongruent, whereas six compar-

TABLE 5. Partitioned Bremer support for each of the data partitions, at interspecific, specific, and subspecific levels. PBS was also correctedby the minimum possible length (mL) of each separate partion.

EF-1α Enolase 16S COI Morphology

PBS /mL PBS /mL PBS /mL PBS /mL PBS /mL

Interspecies 102.4 0.119 128.9 0.109 61.6 0.100 10.3 0.008 5.8 0.105Species 267.0 0.311 299.8 0.253 194.5 0.315 387.0 0.293 7.7 0.140Subspecies 9.5 0.011 33.5 0.028 39.5 0.064 80.0 0.061 −1.5 −0.027Total 378.9 0.442 462.2 0.390 295.6 0.478 477.3 0.362 12.0 0.218

isons that were incongruent in the parsimony analyseswere insignificantly different under likelihood optimiza-tion. The most striking difference was observed in com-parisons involving 16S data that were congruent withall partitions except COI. With parameters optimizedfor the COI data, this partition was incongruent whenforced upon all other topologies, except mtDNA. Fur-thermore, the EF-1α topology was also much more in-congruent in the SH analyses, now incongruent with alldata combined, mtDNA, Enolase and COI, but not 16Sand nucDNA.

Bayesian analysis.—The implementation of a mixedmodel that included separate estimation of parametersfor each coding position, intron, rDNA, and morphol-ogy (12 partitions) produced an identical topology to theML (molecular data only) and Bayesian analyses undera uniform model (Fig. 8). When models were fitted toeach gene and morphological data (5 partitions), three


AS on M


Dow

nloaded from



FIGURE 5. Distribution of partitioned Bremer support values in re-lation to the distance measured from the node tips to the root (all molec-ular data, clock enforced). Stippled line marks the transition betweenpositive and negative contribution to the combined data topology.

shallow nodes shifted their position within clade G (seeFig. 1), moving A. affine and A. euphorbiae to the mostbasal position in that clade, followed by A. jubae.

Speciation Rates

Likelihood ratio tests showed a significant depar-ture from a clock-like substitution rate in the combined

TABLE 6. Maximum-likelihood parameters estimated by Modeltest and selected by the Akaike information criterion (AIC). All base frequen-cies are empirical.

Parameter EF-1α Enolase nDNA 16S COI mtDNA Molecular

Model GTR+I+� GTR+I+� TrN+I+� TIM+I+� TIM+I+� TIM+I+� GTR+I+�

−ln 5562.961 6458.344 12209.649 3376.833 5991.715 9543.586 22211.869pA 0.288 0.316 0.300 0.363 0.302 0.331 0.311pC 0.217 0.197 0.209 0.189 0.227 0.209 0.209pG 0.218 0.240 0.227 0.104 0.168 0.138 0.194pT 0.276 0.247 0.264 0.344 0.303 0.322 0.286r-AC 1.457 1.345 1.000 1.000 1.000 1.000 1.381r-AG 4.846 3.776 3.680 17.724 22.760 19.417 6.098r-AT 2.478 0.793 1.000 1.003 1.621 1.564 1.608r-CG 1.125 1.228 1.000 1.003 1.621 1.564 1.334r-CT 10.539 6.710 6.031 5.545 13.698 9.858 8.229r-GT 1.000 1.000 1.000 1.000 1.000 1.000 1.000Pinv 0.414 0.518 0.534 0.515 0.491 0.526 0.540α 0.490 1.245 0.947 0.501 0.534 0.600 0.814

molecular data as well as for the mtDNA data for 54 taxa.With the outgroups Cisurgus and Crypturgus prunedfrom the mtDNA matrix, however, a clock-like rate couldnot be rejected at the 1% significance level (2lnLR = 73.02,df = 49, P = 0.023). Further deletion of the three pu-tative COI pseudogene sequences (A. aeoni [1], Co. al-laudi [2]) resulted in even more clock-like rates (2lnLR= 64.66, df = 46, P = 0.04). Hence, we used the mtDNAdata with clock-like rates to estimate distances from thetips to the root of the similarly pruned combined dataparsimony tree.

We plotted the cumulative number of lineages of Apha-narthrum and Coleobothrus on the relative time elapsedsince the most recent common ancestor for all speciesof these genera (Fig. 9). This plot shows a more or lessstraight line through time, indicating a nearly constantspeciation to extinction rate.

Geographical Patterns and Host Plant Use

Island colonization.—Within Aphanarthrum andColeobothrus combined, only one unambiguous colo-nization of the islands was found using the “chart statechanges” option in MacClade (accelerated and delayedtransformations indifferent). Five unambiguous backcolorizations to Africa were found on the MP topology(Fig. 10), and three on the ML topology. The alternativeplacements of A. orientalis and A. mairei did not influencethese results. However, when A. armatum was forcedto be monomorphic African, implying recent dispersalto the Canary islands in this species, the numbers ofunambiguous back colorizations were reversed toone for each topology, with up to five possible islandcolorizations (one unambiguous). A maximum of seventransitions were required by the current data, at leastthree of which were recent intraspecific dispersals.

Likelihood estimates of ancestral distributions(African mainland versus islands) supported the gen-eral trends found in the parsimony optimizations. Theancestor of Aphanarthrum was equally likely African orisland distributed. However, all remaining supraspeciesnodes supported island ancestry, involving at least sixback colorizations to the mainland (Fig. 11). When pre-dominantly African distributed species were assumed


AS on M


Dow

nloaded from



FIGURE 6. The most likely tree topologies estimated from each gene partition, using the best model selected by Modeltest (Table 6). SeeFigure 3 for further details.


AS on M


Dow

nloaded from



FIGURE 7. Comparison of the most parsimonious mtDNA and nucDNA consensus trees. Bootstrap support higher than 50% is given aboveeach node. See Table 2 and Figure 3 for further details.

originally African (excluding island distributions), theresult indicated a putative minimum of two islandcolorizations and three back colorizations. Under allanalyses, the island-mainland distributions observed inA. bicinctum and A. affine are most likely due to backcolonization to the continent.

Host switching.—The transition to breeding in Euphor-bia occurred twice in Crypturgini (Fig. 10). Within Apha-narthrum and Coleobothrus, a single origin of breedingin succulents occurred after the transition to Euphor-bia. Likelihood estimation of the ancestral character statein Aphanarthrum suggested that feeding on arborescentshrubs was more than twice as likely as feeding on suc-culent euphorbs (Fig. 10). The number of transitions to

TABLE 7. Percentage increase in parsimony tree lengths when optimized on competing topologies from other partitions or combinationsthereof. Bold numbers mark significant difference as measured by the Wilcoxon signed rank test (Bonferroni corrected significance levels for 49comparisons, 0.001).

Topology→ Data↓ All mtDNA nDNA EF-1α Enolase 16S COI Morph

All — 1.65 0.32 0.92 0.78 1.41 6.83 35.86mtDNA 1.02 — 1.88 2.34 2.44 0.71 1.93 39.92nDNA 0.19 4.13 — 0.68 0.39 3.21 12.20 35.15EF-1α 1.17 5.24 1.05 — 2.91 2.80 16.20 39.51Enolase 0.93 4.82 0.68 2.62 — 4.74 10.91 33.9316S 1.78 0.49 2.91 4.53 4.21 — 11.17 42.56COI 3.03 1.67 3.79 3.64 4.02 3.41 — 40.45Morph 20.00 29.09 20.00 16.36 20.00 23.64 49.09 —

E. balsamifera (one to three) and E. lamarckii complex (oneto two) was more uncertain, with one and three originsof E. lamarckii and E. balsamifera feeding, respectively,under ACCTRAN optimization. Likelihood estimationof ancestral states in the shrub feeding clade also sup-ported three separate origins of E. balsamifera, with allsupraspecies nodes obtaining at least twice the proba-bility for E. lamarckii as the ancestral host in the shrubfeeding clade.

Speciation modes.—Comparisons of sister subspecies,species, and higher groups identified five cases of al-lopatric splits in five subspecies pairs, four allopatric andthree sympatric splits in seven species pairs, and threeallopatric and nine sympatric splits among higher level


AS on M


Dow

nloaded from



FIGURE 8. Comparison of the most parsimonious (left) and Bayesian/ML (right) tree topologies resulting from the analyses of all moleculardata (12-parameter mixed model and uniform model indifferent). Addition of morphological data did not change these topologies in theparsimony or mixed-model Bayesian analyses. Posterior probabilities above 95% are indicated by a star in the Bayesian topology (12 parameters).The line drawing illustrates Aphanarthrum armatum.

clades (Table 8). Only one shift in host plant group wasfound in the sister species or subspecies comparisons,but this event occurred in allopatry (A. euphorbiae).

DISCUSSION

Phylogenetic Analyses

Data combinability.—The simultaneous analysis of fivedifferent data partitions resulted in a well-resolved tree

TABLE 8. Summary of putative speciation modes associated withcladogenetic events among subspecies, species and higher clades.

Geography

Allopatry Sympatry

Host group Same Shift Same Shift

Subspecies 5 0 0 0Species 3 1 3 0Clades 3 0 5 4Subtotal 11 1 8 4Total 12 12

with most nodes receiving higher branch support than inany of the separate analyses. Some of the improvementin branch support was likely due to the increased num-ber of informative characters in the combined data ma-trix. Variable-length bootstrap resampling of each parti-tion showed an overall (but modest) correlation betweenthe number of resampled characters and branch sup-port for nodes congruent with the combined data (ex-cept COI), a pattern observed in numerous other studies(e.g., Cummings et al., 1995; Springer et al., 2001; Koepfliand Wayne, 2003; Rokas et al., 2003). However, the choiceof character sets sampled is not trivial in resolving con-flicting nodes, due to apparent idiosyncrasies of singlegenes. Location dependent processes in the genome, in-cluding biased base frequencies as observed in the COIand Enolase genes (see Table 4), may allow individualpartitions to converge towards wrong topologies (e.g.,Cummings et al., 1995; Mitchell et al., 2000). Samplingcharacters from many independent sources will mostlikely reduce the negative effect of nonindependencewhen location specific idiosyncrasies are diluted in com-bined analyses. This could be a possible explanation for


AS on M


Dow

nloaded from



FIGURE 9. Lineage through time plot showing nearly constant spe-ciation rate in Aphanarthrum. Splits between subspecies were included,but not splits between other populations. Length from the root wascalculated as the number of mitochondrial substitutions per site, usingthe best mtDNA likelihood model with clock enforced.

the largely increased support in the combined data ob-served for the Coleobothrus clade and the clades F and G(cf. Fig. 1) that were incongruent in some of the separateanalyses. Empirical evidence so far have unequivocallyshown that an equal number of characters sampled fromsmaller fragments throughout the genomes resolve thespecies tree more quickly than in a gene by gene proce-dure (e.g. Cummings, 1995; Rokas et al., 2003). Our sam-ple of four gene fragments and morphological data maynot yet constitute a balanced representation of the en-tire genome, explaining why all nodes are not yet fullyresolved. Nevertheless, our results illustrate well howthe process of combining multiple sources of data leadsto complementary support in various hierarchical levelsof the tree topology, and falls in line with several otherstudies which have emphasized the importance of us-ing multiple and independent sources of phylogeneticcharacters (Baker and DeSalle, 1997; Mitchell et al., 2000;Baker et al., 2001; Cameron and Mardulyn, 2001; Cog-nato and Vogler, 2001; Koepfli and Wayne, 2003; Rokaset al., 2003).

Although topological incongruence with the com-bined data ranged from 1 (16S) to 14 (COI) conflictingnodes in the separate parsimony analyses, these conflict-ing nodes had low, if any, branch support. Hence, theanalyses of separate data did not help to clarify any ofthe less resolved relationships from the combined anal-yses; neither did any of the separate gene trees sug-gest any strongly supported alternative hypotheses. Still,measures of incongruence (ILD, Wilcoxon sign rank test,SH test) indicated significant differences among parti-tions, especially between nuclear and mitochondrial par-titions. A general weakness of these tests is that they donot consider the nature of incongruence, such that onerogue taxon may cause the entire data set to be incon-gruent. Also, the likelihood- and parsimony-based in-congruence tests produced very different results, further

reducing confidence in such methods. The use of parti-tioned branch support (partitioned Bremer support sensuBaker, 1997), on the other hand, is a very useful tool inthe localization and quantification of node conflicts. Al-though the significance of Bremer’s (1994) node decayindex is dependent on the branch length (DeBry, 2001),the way that the PBS measure is used here only considersthe relative contribution (or conflict) to each node sepa-rately. The largest cost of any data partitions enforced onthe combined topology was four extra steps in one Eno-lase node (see Fig. 5), which is nevertheless insignificantwhen branch length is taken into account (cf. Table 3 inDeBry, 2001). The heterogeneity measured between thenuclear and mitochondrial partitions was hardly visiblein the PBS analyses, with only three nodes having a neg-ative contribution from both mtDNA partitions (cladesincluding A. affine, A. euphorbiae, A. jubae; A. hesperidum,A. piscatorium; all of these plus A. bicolor). However, thesethree nodes were characterized by very short ML branchlengths in separate and combined analyses (cf. Figs. 6,8), and the apparent conflict may reflect random sortingevents during a short period of elevated speciation rates(cf. Fig. 9).

Potential negative effects of combining ‘statistically’incongruent partitions involve less resolved or less sup-ported tree topologies. However, exclusion or sequentialdown weighting of the most incongruent partitions, COI(entire gene fragment or third positions only) or mor-phology, did not change the combined data topologies inneither of the parsimony or Bayesian analyses. Furtherexclusion of alignment ambiguous regions still resultedin the same topology, with node support remaining es-sentially the same, or slightly lower, through all theseanalyses. Moreover, the application of a mixed model inthe Bayesian analyses did not produce different resultsfrom the uniform Bayesian analysis. Hence, the Bayesianresults suggest that the effect of fitting separate modelsto subpartitions become less critical with an increasednumber of independent character sources.

Overall, our data have demonstrated how significantmeasures of incongruence can mislead when conflict infact is very low (Baker et al., 2001). The incongruenceobserved between the separate data partitions then pos-sibly reflects the smaller number of informative charac-ters in each of them, which makes tree estimation moresusceptible to homoplasious changes, in particular thosepartitions experiencing the highest substitution rates.

Phylogenetic relationships.—This study clearly demon-strated the paraphyly of Aphanarthrum with respect toColeobothrus. The basal position of A. armatum was sup-ported by all data sets except COI, in either parsimony orlikelihood analyses, or both. COI performed much worsethan any other partition, and had the highest level ofhomoplasy, significant nonstationarity in base composi-tion, the fewest supported nodes, and the lowest branchsupport per informative character. The quite differentproperties (and congruence with the combined data tree)of the completely linked 16S data emphasize that thepoor performance of the COI data is not genome spe-cific, and confirm previous findings that COI is not very


AS on M


Dow

nloaded from



FIGURE 10. Parsimony (branches) and likelihood (pies) reconstruction of host plant use on the combined parsimony topology. Succulentsinclude E. canariense, E. handiense, E. echinus, E. teke; the ‘E. lamarckii’ complex includes E. regis-jubae, E. lamarckii, E. artropurpurea, E. tuckeyana,E. piscatoria. Pies show the probability of ancestral host plant groups (succulents versus arborescent shrubs). The right hand matrix displays thegeographical distribution of each species or subspecies: V, Cape Verde archipelago; M, Madeira archipelago; H, El Hierro; P, La Palma; G, LaGomera; T, Tenerife; C, Gran Canaria; F, Fuerteventura; L, Lanzarote (all Canary Islands); Aw, western Africa (Morocco, Senegal, and Gambia);Ae, eastern Africa (Uganda, Kenya, Sudan, Ethiopia); X, Holarctic region.


AS on M


Dow

nloaded from



FIGURE 11. Likelihood estimation of the geographical distribution of ancestors (mainland, black; Macaronesian islands, white). Terminal boxesindicate the current distribution. The relative proportion in each pie indicates the probability of each ancestral character state. The alternativereconstruction shown on the right assumes recent African to Macaronesian dispersal in some species, including the basal A. armatum.

useful as a stand alone data source for higher level insectphylogenetics (Durando et al., 2000; Baker et al., 2001;Jordal et al., 2002; Danforth et al., 2003). However, itis also possible that our COI data included conservedmitochondrial pseudogenes integrated in the nucleargenome. Although none of these sequences containedstop codons, several recent studies have detected multi-ple integrations of indel-free mitochondrial pseudogenes(Bensasson et al., 2001, but see Williams and Knowlton,2001, for an extreme example). Until more conclusiveevidence is obtained in favor or disfavor of the pseudo-gene hypothesis, we caution against any strong interpre-tations of the COI data. We note, however, that exclusionof the COI data did not change the topology, or the av-erage bootstrap support, rendering these data harmlessin the face of potential paralogy.

The inclusion of A. mairei in the Coleobothrus clade, andthe basal position of A. orientalis (ML/Bayesian) in thisclade, reinforced the evidence against the monophylyof Coleobothrus. Although the two Aphanarthrum specieshad base compositions more similar to the Coleoboth-

rus species than to the other Aphanarthrum species,LogDet distance analyses of the nuclear DNA data, orall molecular data, did not shift the position of thesespecies. Thus, biased base frequencies can not aloneexplain these relationships (DeBry, 2003). Also severalsuits of morphological characters associate A. orientalisand A. mairei, with Coleobothrus (Fig. 2). Male genitaliaare nearly identical in these five species, all with ahair-pinned ventral sclerite and asymmetrical tegmen(Appendices 1, 2). Consequently, Coleobothrus must betreated as a synonym of Aphanarthrum, thus rejectingMenier’s proposed generic status of Coleobothrus (Me-nier, 1973). In the other taxa of the succulent breeders, A.canariense and A. pygmeum, genitalia are highly special-ized, or reduced, showing how rapid genitalic characterscan diverge in secondarily sympatric species. However,molecular data left no doubt about the monophyly of allspecies breeding in succulent euphorbs.

The 12 remaining Macaronesian Aphanarthrum, whichall breed in shrubs, made up one strongly supportedclade. The support for A. bicinctum and A. canescens as


AS on M


Dow

nloaded from



the two most basal taxa was also strong. Even thoughthe separate analyses disagreed upon which of thesetaxa should be most basal, most of them, as well as thecombined parsimony, likelihood and Bayesian analyses,placed A. bicinctum in that position. All 12 species areexternally very similar to each other, but the shape of thetransverse elytral fasciae distinguish the two basal taxafrom the remaining ten. Whereas most nodes in this cladewere strongly supported, two short internodes had thelowest possible branch support (two basal nodes in A.glabrum’s sister group; Fig. 1). The two nodes were alsothose that most frequently changed under various com-binations of partitions (and various optimality criteria)and we consider these unresolved until more data arecollected. However, it seems a difficult task to achievewell supported resolutions for these nodes, based onthe low branch support provided by all five parti-tions. The low resolution was also reflected in shortinternal branch lengths in each of the underlyinggene trees (cf. Fig. 6), corresponding to a small butsudden increase in speciation rate (cf. Fig. 9). Un-der such circumstances, only large amounts of datawould provide a sufficient number of informative char-acters needed to improve phylogenetic accuracy andconfidence.

Given the extensive overlap and general lack of gooddiagnostic external characters, the strongly supportedmonophyly of each species was strikingly clear. Withonly one exception, in the paraphyletic glabrum complex,these species have existed as separate evolutionary en-tities for millions of years (all interspecific COI diver-gences above 10%). The deep coalescence between eco-logically and morphologically similar species thus refutethe hypotheses that hybridization or recent diversifica-tion act as possible confusing factors in the evolution ofmorphological variation.

Subspecies Formation

The paraphyly of A. glabrum with respect to A. sub-glabrum, and the very little genetic divergence betweenthe two species, illustrate a case of very recent specia-tion. The paraphyletic signature is a predicted one un-der circumstances where new species originate by pe-ripheral isolation, and does not preclude species statusof the two species involved (Harrison, 1998). Some spe-ciation events must necessarily be recent and too littletime has elapsed for complete and reciprocal monophylyto evolve for every single gene tree. Population sizesin these species, as in most species of Aphanarthrum,are enormous and probably require millions of gener-ations to obtain exclusive groups. Nevertheless, the Eno-lase gene strongly supported reciprocal monophyly ofeach of the two glabrum subspecies and A. subglabrum,suggesting that selection on certain nuclear genes mayfollow the assumed reproductive isolation between thetwo species on La Palma. More samples are needed toexplore the exact nature of this incipient speciation pro-cess, and a more detailed investigation is currently un-der way. What this example has clearly demonstrated

though is the need for sampling multiple populationsand markers to discover the paraphyletic nature ofspecies.

Sampling of multiple populations also provided infor-mation on the reality of proposed subspecies (Israelson,1972). Overall, most species showed clear geographicaldivergences corresponding to the suggested subspecificclassification. Although we did not manage to sampleone of the A. canescens subspecies, and A. bicinctum obsi-tum from Gran Canaria and A. jubae tuberculatum couldnot be judged by the limited sample included, all othersubspecies were clearly distinct and each of the gene treesshowed reciprocal monophyly of A. canescens and A. ca-nariense subspecies, whereas Enolase and the combineddata showed the same for A. bicinctum. Very deep co-alescence characterized the first two subspecies splits,which were even deeper than some of the splits betweenmorphologically distinct species (e.g., A. bicolor versusA. hesperidum versus A. piscatorium, or A. wollastoni ver-sus A. aeoni). Taken together with the almost straight lineLTT plot (Fig. 9), the recent formation of subspecific lin-eages fits the general picture of a group undergoing ac-tive evolution and continuous speciation.

Factors Associated with Speciation

Host plant switching has only played a minor role inrecent speciation events in crypturgine beetles. Amongthe total of 24 cladogenetic events in the Aphanarthrumclade (Fig. 10), only 6 of these were associated with ma-jor changes in host plant use, significantly lower thanexpected from a random distribution of character states(8 to 12, average 10.3). The use of succulent euphorbswas particularly conservative, with no reversals to otherhost groups within Aphanarthrum. Given that this lineagehas existed for a long time, and that both succulents andshrub-like euphorbs coexist in most Canary islands andin Morocco, stasis in host plant use may be interpretedas strong constraints on host switching. Switching be-tween the various shrub-like euphorb groups was morefrequent, but even the most recent of these (to E. longi-folia) involved considerable sequence divergence (12.8%to 13.5% COI, uncorrected), and occurred in allopatry.

Many recent sister lineages are allopatrically dis-tributed, suggesting that these diverged by geographi-cal isolation on different islands. The degree of sympatryincreases with increasing phylogenetic depth, however,suggestive of secondary range expansion (Barracloughet al., 1998; Barraclough and Vogler, 2000). Although al-lopatric differentiation may be the most universal specia-tion mode in island archipelagos, as has also been shownin some other insect groups (Jordan et al., 2003; Percy,2003), several studies have reported a more even contri-bution (Kambysellis and Craddock, 1997; Joy and Conn,2001), or even favored ecological specialization as themost prominent isolating factor on islands (Shaw, 2002).

Colonization Patterns

Several lines of evidence suggest that the relativelyshort distance between the African continent and the


AS on M


Dow

nloaded from



Atlantic islands has not been an important barrier to dis-persal in this group of insects. Contrary to the single-island colorizations seen in many flightless beetles (Juanet al., 2000; Gillespie and Roderick, 2002), the high fre-quency of geographic interchange in Aphanarthrum bee-tles seems connected to their flight capacity. Recent ex-change has taken place within A. mairei, A. affine, andA. bicinctum, and seems equally likely in two morespecies, A. armatum and A. allaudi, which have previouslybeen collected in western Africa and Fuerteventura, re-spectively (Fig. 10). Although the most parsimonious orlikely solution of African-Macaronesian exchange sug-gests a Macaronesian origin of Aphanarthrum and fiveback colonisations to the mainland, this is dependenton the ancestral distribution of the basal A. armatum.Given the frequent interchange and high dispersal ca-pacity in these beetles, A. armatum could have originallybeen restricted to Africa only, implying up to three an-cient colonisations of Macaronesia. A reasonable consen-sus area of origin may include the Moroccan coast andthe oldest (eastern) Canary islands. After all, the mostbasal taxon is currently found in this area, and the an-cestor of at least one of the eastern African species, Co.germeauxi, most likely occurred in the same area.

Whereas the geographical origin of Aphanarthrum re-mains uncertain, the monophyletic clade of Macarone-sian shrub breeding species implies a single commonancestor from the Canary Islands. Furthermore, the twospecies endemic to the Cape Verde and Madeira islandsare nested within this clade, related to species endemic toa restricted number of islands. Aphanarthrum hesperidumis related to A. piscatorium and A. bicolor, a group ofmainly western Canary Islands species. Interestingly, thehost plant in Cape Verde, E. tuckeyana, is closely relatedto species in the lamarckii complex in the Canary Islandsas well (Molero et al., 2002), and both host plant andbeetle species seem to have followed the same disper-sal route. A similar hypothesis exists for one clade ofLiparthrum beetles that has the same specialized asso-ciation with Euphorbia (Jordal et al., 2003). Little moreis known about the origin of the Cape Verde flora andfauna, but at least in Echium plants there is strong ev-idence in favour of an eastern Canary Islands to CapeVerde dispersal route. The species endemic to Madeira,A. euphorbiae, is related to an eastern Canary Island plusMoroccan species, but is nested within a clade of CanaryIsland endemics, suggesting that these islands served asthe ancestral area. Recent dispersal in A. piscatorium andA. bicolor from the western islands, in particular Tenerife(phylogeographical studies in progress, but see Fig. 1), toseveral Madeiran islands further supports our hypothe-sis for a Canary Island to Madeira dispersal route. Rel-atively more is known about the origin of the Madeiranflora and fauna, which have been derived from a varietyof geographical sources. Among these, only plants seemto have colonized from the western islands (Bohle et al.,1996; Panero et al., 1999; Barber et al., 2002), whereas sev-eral beetle groups had a more eastern origin (Emersonet al., 2000; Rees et al., 2001), and examples of trees, but-terflies, and reptiles indicate an Iberian or Moroccan con-

tinental origin (Brunton and Hurst, 1998; Nogales et al.,1998; Hess et al., 2000).

CONCLUSION

This phylogenetic study has demonstrated the impor-tance of thorough sampling in two different ways. First, itshows how samples from multiple populations can vali-date or reject the existence of genetically distinct subspe-cific lineages, which can further provide sufficient evi-dence to assess the phyletic status of species. Secondly, ithighlights the many pitfalls associated with limited sam-pling of characters. Whereas most character partitionsproduced many identical results, no single partition hadthe capacity of resolving all nodes with confidence, orachieve congruence with a topology based on all avail-able data. Because node-specific conflict was low amongthe partitions, incongruence was probably due to differ-ent levels of homoplasy, and not necessarily to differentgenealogical histories, further supporting the strategy ofcombining all data to estimate a new hypothesis on Apha-narthrum phylogeny.

The relatively low frequency of host switching rendersgeographical isolation a much more important factor inthe recent diversification of Aphanarthrum beetles. How-ever, rare transitions into novel resources have never-theless opened avenues of new opportunities for thesebeetles, by providing unexploited ecological niches forfurther diversification. Thus, the contribution from onemajor historical host switch, e.g., to succulent growthforms, has provided the Aphanarthrum lineage with 10additional species. This demonstrates in the simplestpossible way how important host switching may be tothe total diversity, even though other factors seem cur-rently much more influential.

ACKNOWLEDGEMENTS

The potential of Macaronesian Aphanarthrum as a useful evolutionarymodel system was first pointed out by Lawrence Kirkendall and HaraldBreilid. Together with Kjetil Harkestad they also provided specimensduring the initial stage of the project. We would further like to thankPedro Oromi and Roberto Jardim for help with logistics and permits tocollect in some of the national parks; many thanks also to each of the‘Cabildos’ in the Canary Islands for issuing collecting permits. SharonShute and Martin Baehr kindly arranged access to type material in theNatural History Museum, London, and Zoologische Staatssammlung,Munich. Finally, we wish to thank Chris Simon, Patrick Mardulyn,and Mike Caterino for helpful comments on a previous version of thispaper. This project was funded by a Marie Curie Fellowship HPMF-CT2001-01323 to BHJ.

REFERENCES

Baker, R. H., and R. DeSalle. 1997. Multiple sources of character in-formation and the phylogeny of Hawaiian drosophilids. Syst. Biol.46:645–673.

Baker, R. H., G. S. Wilkinson, and R. DeSalle. 2001. Phylogenetic util-ity of different types of molecular data used to infer evolutionaryrelationships among stalk-eyed flies (Diopsidae). Syst. Biol. 50:87–105.

Barber, J. C., J. Francisco-Ortega, A. Santos-Guerra, K. G. Turner, andR. K. Jansen. 2002. Origin of Macaronesian Sideritis L. (Lamioideae:Lamiaceae) inferred from nuclear and chloroplast sequence datasets.Mol. Phylogenet. Evol. 23:293–306.


AS on M


Dow

nloaded from



Barker, F. K., and F. M. Lutzoni. 2002. The utility of the IncongruenceLength Difference test. Syst. Biol. 51:625–637.

Barraclough, T. G., and S. Nee. 2001. Phylogenetics and speciation.Trends Ecol. Evol. 16:391–399.

Barraclough, T. G., and A. P. Vogler. 2000. Detecting the geographi-cal pattern of speciation from species-level phylogenies. Am. Nat.155:419–434.

Barraclough, T. G., A. P. Vogler, and P. H. Harvey. 1998. Revealing thefactors that promote speciation. Phil. Trans. R. Soc. Lond. B 353:241–249.

Bensasson, D., D. Zhang, D. L. Hartl, and G. M. Hewitt. 2001. Mi-tochondrial pseudogenes: Evolution’s misplaced witnesses. TrendsEcol. Evol. 16:314–321.

Bohle, U., H. H. Hilger, and W. F. Martin. 1996. Island colonization andevolution of the insular woody habit in Echium L. (Boraginaceae).Proc. Natel. Acad. Sci. U.S.A. 93:11740–11745.

Bremer, K. 1994. Branch support and tree stability. Cladistics 10:295–304.

Brown, R. P., N. M. Suarez, A. Smith, and J. Pestano. 2001. Phylogeog-raphy of Cape Verde Island skinks (Mabuya). Mol. Ecol. 10:1593–1597.

Brunton, C. F. A., and G. D. D. Hurst. 1998. Mitochondrial DNAphylogeny of Brimstone butterflies from the Canary Islands andMadeira. Biol. J. Linn. Soc. 63:69–79.

Buckley, T. R., C. Simon, and G. K. Chambers. 2001. Exploring among-site rate variation models in a maximum likelihood framework us-ing empirical data: Effects of model assumptions on estimates oftopology, branch lengths, and bootstrap support. Syst. Biol. 50:67–86.

Buckley, T. R., C. Simon, P. K. Flook, and B. Misof. 2000. Secondarystructure and conserved motifs of the frequently sequenced domainsIV and V of the insect mitochondrial large subunit rRNA gene. InsectMol. Biol. 9:565–580.

Cameron, S. A., and P. Mardulyn. 2001. Multiple molecular data setssuggest independent origins of highly eusocial behaviour in bees(Hymenoptera: Apinae). Syst. Biol. 50:194–214.

Carter, S., and A. R. Smith. 1988. Euphorbiaceae (part 2). Flora of Trop-ical East Africa (R. M. Polhill, ed.). A. A. Balkema, Rotterdam.

Cognato, A. I., and A. P. Vogler. 2001. Exploring data interaction andnucleotide alignment in a multiple gene analysis of Ips (Coleoptera:Scolytinae). Syst. Biol. 50:758–780.

Cummings, M. P., S. P. Otto, and J. Wakeley. 1995. Sampling proper-ties of DNA sequence data in phylogenetic analysis. Mol. Biol. Evol.12:814–822.

Danforth, B. N., L. Conway, and S. Ji. 2003. Phylogeny of eusocial La-sioglossum reveals multiple losses of eusociality within a primitivelyeusocial clade of bees (Hymenoptera: Halictidae). Syst. Biol. 52:23–36.

DeBry, R.W., 2001. Improving interpretation of the Decay Index forDNA sequence data. Syst. Biol. 50:742–752.

DeBry, R. W. 2003. Identifying conflicting signal in a multigene anal-ysis reveals a highly resolved tree: The phylogeny of rodentia(Mammalia). Syst. Biol. 52: 604–617.

Despres, L., E. Pettex, V. Plaisance, and F. Pompanon. 2002. Speciationin the Globeflower fly Chiastocheta spp. (Diptera: Anthomyiidae) inrelation to host plant species, biogeography, and morphology. Mol.Phylogenet. Evol. 22:258–268.

Durando, C. M., R. H. Baker, W. J. Etges, W. B. Heed, M. Wasserman,and R. DeSalle. 2000. Phylogenetic analysis of the repleta group ofthe genus Drosophila using multiple sources of characters. Mol. Phy-logenet. Evol. 16:296–307.

Emerson, B. C. 2002. Evolution on oceanic islands: Molecular phyloge-netic approaches to understanding pattern and process. Mol. Ecol.11:951–966.

Emerson, B. C., P. Oromi, and G. M. Hewitt. 2000. Interpreting coloniza-tion of the Calathus (Coleoptera: Carabidae) on the Canary Islandsand Madeira through the application of the parametric bootstrap.Evolution 54:2081–2090.

Farrell, B. D., A. Sequeira, B. O’Meara, B. B. Normark, J. Chung,and B. Jordal. 2001. The evolution of agriculture in beetles(Curclionidae: Scolytinae and Platypodinae). Evolution 55:2011–2027.

Farris, J. S., M. Kallersjo, A. G. Kluge, and C. Bult. 1995. Con-structing a significance test for incongruence. Syst. Biol. 44:570–572.

Gillespie, R. G., and G. K. Roderick. 2002. Arthropods on islands:Colonization, speciation, and conservation. Ann. Rev. Ent. 47:595–632.

Harrison, R. H. 1998. Linking evolutionary pattern and process: Therelevance of species concepts for the study of speciation. Pages 19–31 in Endless forms: Species and speciation (D. J. Howard and S. H.Berlocher, eds.). Oxford University Press, Oxford.

Hess, J., J. W. Kadereit, and P. Vargas. 2000. The colonization historyof Olea europaea L. in Macaronesia based on internal transcribedspacer 1 (ITS-1) sequences, randomly amplified polymorphic DNAs(RAPD), and intersimple sequence repeats (ISSR). Mol. Ecol. 9:857–868.

Israelson, G. 1972. Male copulatory organs of Macaronesian speciesof Aphanarthrum Wollaston. With designations of lectotypes anddescriptions of new taxa (Col. Scolytidae). Ent. Scand. 3:249–257.

Israelson, G. 1976. Redescription of Deropria elongata (Eggers), withnotes on some species of Aphanarthrum Wollaston (Coleoptera,Scolytidae). Zool. Mededelingen 50:39–44.

Israelson, G. 1979. Taxonomical and nomenclatural notes on some Ca-narian Coleoptera. Vieraea 9:183–210.

Jordal, B. H. 2002. Elongation Factor 1α resolves the monophyly of thehaplodiploid ambrosia beetles Xyleborini (Coleoptera: Curculion-idae). Insect Mol. Biol. 11:453–465.

Jordal, B. H., L. R. Kirkendall, and K. Harkestad. 2004. Phylogeny ofa Macaronesian radiation: Host-plant use and possible cryptic spe-ciation in Liparthrum bark beetles. Mol. Phylogenet. Evol. 31:554–571.

Jordal, B. H., B. B. Normark, B. D. Farrell, and L. R. Kirkendall. 2002.Extraordinary haplotype diversity in haplodiploid inbreeders: Phy-logenetics and evolution of the sib-mating bark beetle genus Coc-cotrypes. Mol. Phylogenet. Evol. 23:171–188.

Jordan, S., C. Simon, and D. Polhemus. 2003. Molecular systematics andadaptive radiation of Hawaii’s endemic damselfly genus Megalagrion(Odonata: Coenagrionidae). Syst. Biol. 52:89–109.

Joy, D. A., and J. E. Conn. 2001. Molecular and morphological phyloge-netic analysis of an insular radiation in Pacific black flies (Simulium).Syst. Biol. 50:18–38.

Juan, C., B. C. Emerson, P. Oromi, and G. M. Hewitt. 2000. Colonizationand diversification: Towards a phylogeographic synthesis for theCanary Islands. Trends Ecol. Evol. 15:104–109.

Kambysellis, M. P., and E. M. Craddock. 1997. Ecological and reproduc-tive shifts in the diversification of the endemic Hawaiian Drosophila.Pages 475–509 in Molecular evolution and adaptive radiation (T.J. Givnish and K. J. Sytsma, eds.). Cambridge University Press,Cambridge.

Kelley, S. T., and B. D. Farrell. 1998. Is specialization a dead end? Thephylogeny of host use in Dendroctonus bark beetles (Scolytidae). Evo-lution 52: 1731–1743.

Koepfli, K. and R. K. Wayne. 2003. Type I STS markers are moreinformative than Cytochrome b in phylogenetic reconstructionof the Mustelidae (Mammalia: Carnivora). Syst. Biol. 52: 571–593.

Lee, M. S. Y. 2001. Unalignable sequences and molecular evolution.Trends Ecol. Evol. 16:681–685.

Lockhart, P. J., M.A. Steel, M.D. Hendy, and D. Penny. 1994. Recov-ering evolutionary trees under a more realistic model of sequenceevolution. Mol. Biol. Evol. 11: 605–612.

Losos, J. B., and R. E. Glor. 2003. Phylogenetic comparative methodsand the geography of speciation. Trends Ecol. Evol. 18:2003.

Machado, C. A., E. Jousselin, F. Kjellberg, S. G. Compton, and E. A.Herre. 2000. Phylogenetic relationships, historical biogeography andcharacter evolution of fig-pollinating wasps. Proc. R. Soc. LondonSer. B 268:685–694.

Maddison, W. P., and D. R. Maddison, 1992. MacClade. Version 3.04.Sinauer Associates, Sunderlands, Massachusetts.

Menier, J. J. 1973. Rehabilitation du genre Coleobothrus Enderlein etdescription d’une espece nouvelle de l’est africain: C. germeauxi. Bull.Soc. Entomol. France 78:205–209.


AS on M


Dow

nloaded from



Mitchell, A., C. Mitter, and J. C., Regier. 2000. More taxa or more charac-ters revisited: Combining data from nuclear protein-encoding genesfor phylogenetic analyses of Noctuoidea (Insecta: Lepidoptera). Syst.Biol. 49:202–224.

Molero, J., T. Garnatje, A. Rovira, N. Garcia-Jacas, and A. Susanna. 2002.Karyological evolution and molecular phylogeny in Macaronesiandendroid spurges (Euphorbia subsect. Pachycladae). Plant Syst. Evol.231:109–132.

Nixon, K. C., and J. M. Carpenter. 1996. On simultaneous analysis.Cladistics 12:221–241.

Nogales, M., M. Lopez, J. Jimenez-Asensio, J. M. Larruga, M.Hernandez, and P. Gonzalez. 1998. Evolution and biogeographyof the genus Tarentola (Sauria: Gekkonidae) in the Canary Islands,inferred from mitochondrial DNA sequences. J. Evol. Biol. 11:481–494.

Normark, B. B., B. H. Jordal, and B. D. Farrell. 1999. Origin ofa haplodiploid beetle lineage. Proc. R. Soc. Lond. B 266:2253–2259.

Pagel, M. 1994. Detecting correlated evolution on phylogenies: A gen-eral method for the comparative analysis of discrete characters. Proc.R. Soc. London Ser. B 255:37–45.

Pagel, M. 1999. The maximum likelihood approach to reconstructingancestral character states of discrete characters on phylogenies. Syst.Biol. 48:612–622.

Panero, J. L., F. Francisco-Ortega, R. K. Jansen, and A. Santos-Guerra.1999. Molecular evidence for multiple origins of woodiness and aNew World biogeographic connection of the Macaronesian Islandendemic Pericallis (Asteraceae: Senecioneae). Proc. Natl. Acad. Sci.U. S. A. 96:13886–13891.

Pellmyr, O., and J. Leebens-Mack. 2000. Reversal of mutualism foradaptive radiation in Yucca moths. Am. Nat. 156:S62–S76.

Percy, D. M. 2003. Radiation, diversity and host plant interactionsamong island and continental legume-feeding psyllids. Evolution57:2540–2556.

Poe, S., and D. L. Swofford. 1999. Taxon sampling revisited. Nature398:299–300.

Pollock, D. D., D. J. Zwickl, J. A. McGuire, and D. M. Hillis. 2002. In-creased taxon sampling is advantageous for phylogenetic inference.Syst. Biol. 51:664–671.

Posada, D., and K. A. Crandall. 1998. Modeltest: Testing the model ofDNA substitution. Bioinformatics 14:817–818.

Rees, D. J., B. C. Emerson, P. Oromi, and G. M. Hewitt. 2001. Thediversification of the genus Nesotes (Coleoptera: Tenebrionidae) inthe Canary Islands: Evidence from mtDNA. Mol. Phylogenet. Evol.21:321–326.

Rokas, A., B.L Williams, N. King and S. B. Carroll. 2003. Genome-scaleapproaches to resolving incongruence in molecular phylogenies. Na-ture 425:798–804.

Ronquist, F., and J. P. Huelsenbeck. 2003. MRBAYES 3: Bayesian phy-logenetic inference under mixed models. Bioinformatics 19:1572–1574.

Rosenberg, M. S., and S. Kumar. 2001. Incomplete taxon sampling is nota problem for phylogenetic inference. Proc. Natel. Acad. Sci. U.S.A.98:10751–10756.

Scheffer, S. J., and B. M. Wiegmann. 2000. Molecular phylogeneticsof the holly leafminers (Diptera: Agromyzidae: Phytomyza): Specieslimits, speciation and dietary specialization. Mol. Phylogenet. Evol.17:244–255.

Seigler, D. S. 1994. Phytochemistry and systematics of the Euphor-biaceae. Ann. Missouri Bot. Garden 81:380–401.

Shaw, K. L. 2002. Conflict between nuclear and mitochondrial DNAphylogenies of a recent species radiation: What mtDNA reveals andconceals about modes of speciation in Hawaiian crickets. Proc. Natl.Acad. Sci. U.S.A. 99:16122–16127.

Shimodaira, H., and M. Hasegawa. 1999. Multiple comparisons of log-likelihoods with applications to phylogenetic inference. Mol. Biol.Evol. 16:1114–1116.

Sorenson, M. D. 1999. TreeRot, version 2. Boston University, Boston.Springer, M. S., R.W. DeBry, C. Douady, H. M. Amrine, O. Madsen,

W. W. deJong, and M. J. Stanhope. 2001. Mitochondrial versus nucleargene sequences in deep-level mammalian phylogeny reconstruction.Mol. Biol. Evol. 18:132–143.

Thompson, J. D., T. J. Gibson, F. Plewniak, F. Jeanmougin, and D. G.Higgins. 1997. The ClustalX windows interface: Flexible strategiesfor multiple sequence alignment aided by quality analysis tools. Nu-cleic Acids Res. 24:4876–4882.

Tracy, M. R., and S. B. Hedges. 2000. Evolutionary history of the Enolasegene family. Gene 259:129–138.

Webster, G. L. 1994. Classification of the Euphorbiaceae. Ann. MissouriBot. Garden 81:3–32.

Wheeler, W. C., J. Gatesy, and R. DeSalle. 1995. Elision: A methodfor accomodating multiple molecular sequence alignments withalignment-ambiguous sites. Mol. Phylogenet. Evol. 4:1–9.

Williams, S. T., and N. Knowlton. 2001. Mitochondrial pseudogenes arepervasive and often isidious in the snapping shrimp genus Alphaeus.Mol. Biol. Evol. 18:1484–1493.

Wood, S. L. 1986. A reclassification of the genera of Scolytidae(Coleoptera). Gt. Basin Nat. Mem. 10:126.

Wood, S. L. 1988. Nomenclatural changes and new species of Scolytidae(Coleoptera). Gt. Basin Nat. 48:188–195.

Yoder, A. D., J. A. Irwin, and B. A. Payseur. 2001. Failure of the ILD todetermine data combinability for Slow Loris phylogeny. Syst. Biol.50:408–424.

First submitted 1 June 2003; reviews returned 27 October 2003;final acceptance 14 March 2003

Associate Editor: James Whitfield

APPENDIX 1Morphological characters and description of character states. Aftereach description follows the maximum possible steps for the character,and the consistency (CI) and retention (RI) indices as measured overthe combined data parsimony tree (Fig. 1), but including only one ter-minal per species or subspecies. Characters with a maximum of onestep apply to outgroup apomorphies only.

Male Genitalia

1. Anterior part of spiculum gastrale: (0) simple rod; (1) with tinytooth or nod; (2) Y-shaped fork; (3) L-shaped; (4) with subapicaltooth. Max steps 12, CI = 0.75, RI = 0.89.

2. Apophyses: (0) long and slender, longer than tube; (1) as short astube. Max steps 11, CI = 0.5, RI = 0.90.

3. Tegmen: (0) symmetric; (1) asymmetric. Max steps 5, CI = 1.0,RI = 1.0.

4. Manubrium: (0) long, narrow; (1) small, acute; (2) absent. Max steps7, CI = 0.67, RI = 0.80.

5. Flagellum: (0) short; (1) long, coiled. Max steps 4, CI = 0.50,RI = 0.67.

6. End plates: (0) absent; (1) present. Max steps 7, CI = 0.50, RI = 0.83.7. Dorsal face of end plates: (0) smooth; (1) fringed. Max steps 2,

CI = 1.0, RI = 1.0.8. Spines on end plates: (0) absent; (1) few and scattered; (2) many.

Max steps 3, CI = 0.67, RI = 0.00.9. Lamina (ventral sclerite): (0) absent; (1) present. Max steps 5,

CI = 1.0, RI = 1.0.10. Terminal tip of lamina: (0) unfolded; (1) folded. Max steps 3,

CI = 0.50, RI = 0.50.11. Anterior lamina: (0) broad; (1) thick, U-shaped; (2) thin, U-shaped;

(3) spoon-shaped; (4) narrowly reduced; (5) slender hair-pin; (6)composite. Max steps 19, CI = 0.75, RI = 0.85.

Head

12. Lower frons: (0) smooth; (1) reticulate. Max steps 12, CI = 0.33,RI = 0.82

13. Frons: (0) flat; (1) concave; (2) convex. Max steps 3, CI = 0.50,RI = 0.50.

14. Distance between eyes: (0) 3 × width of eye; (1) 2 × width of eye;(2) 1 × width of eye. Max steps 6, CI = 0.67, RI = 0.75.

15. Eyes: (0) weakly sinuate; (1) emerginate. Max steps 1 (CI, RI n/a).


AS on M


Dow

nloaded from



16. Funicle: (0) 3-segmented; (1) 2-segmented. Max steps 1 (CI, RI n/a).17. Sutures on antennal club: (0) 2, recurved; (1) 2, constricted; (2) 2,

compressed at tip; (3) absent. Max steps 5, CI = 1.0, RI = 1.0.18. Septum of antennal club: (0) absent; (1) single septum; (2) pair at

suture 1. Max steps 5, CI = 0.67, RI =0.67.

Pronotum

19. Pronotal surface: (0) smooth; (1) reticulate. Max steps 7, CI = 0.25,RI = 0.50.

20. Pronotal disc: (0) evenly rounded; (1) summit at basal third;(2) flattened. Max steps 7, CI = 0.67, RI = 0.80.

21. Pronotal shape: (0) broadly rounded; (1) narrowly rounded; (2)constricted at anterior fourth. Max steps 13, CI = 0.67, RI = 0.91.

22. Pronotal tubercles at anterior margin: (0) absent; (1) 2–4; (2) 2–4,margin recurvec; (3) 8–10. Max steps 11, CI = 0.33; RI = 0.56.

23. Pronotal asperities: (0) absent ; (1) fine; (2) coarse, extends to pos-terior parts. Max steps 3, CI = 1.0, RI = 1.0.

Elytra

24. Fasciae on disc: (0) absent; (1) transverse only; (2) with me-dian rings; (3) median rings squared, open anteriorly; (4) me-dian rings filled; (5) median area V-shaped. Max steps 18, CI = 1.0,RI = 1.0.

25. Fasciae on declivity: (0) absent; (1) present. Max steps 13, CI = 0.50,RI = 0.92.

26. Declivity: (0) rounded; (1) concave. Max steps 3, CI = 0.50, RI = 0.50.

Legs

27. Protibial teeth: (0) 10; (1) 6; (2) 5; (3) 4. Max steps 7, CI = 0.33,RI = 0.20.

28. Tarsal segments 1–4: (0) longer than 5; (1) shorter than 5. Max steps1 (CI, RI, n/a).

APPENDIX 2Morphological character matrix for the 28 characters described inAppendix 1.

Dolurgus pumilus 000200- -0- -120100010 00000010Deropria elongata 40000100?? ?1??01301? 03200001Crypturgus borealis 000010- -0- -120112112 20000020Cr. hispidulus 000010- -0- -020112102 20000020Cisurgus wollastoni 000010- -0- -001112002 20000030Ci. occidentalis 000010- -0- -001112002 20000030Coleobothrus luridus 1012010010 5001111200 20000120Co. allaudi 1012010010 5001111200 20000120Co. germeauxi 1012010010 5001111210 20000130Aphanarthrum armatum 000000- -10 5001111210 11011020A. mairei 1012010010 5001111210 20000030A. orientalis 1012010010 5001111210 20000030A. pygmeum 010200- -10 4002111211 22150020A. canariense canariense 2101010010 6002111211 22150030A. canariense neglectum 2101010010 6002111211 22150030A. bicinctum bicinctum 1102010010 1101111210 10031030A. bicinctum obsitum 1102010010 1101111210 10031030A. bicinctum vestitum 1102010010 1101111210 10041030A. canescens canescens 1102011010 1101111210 10041030A. canescens polyspiniger 1102011010 1101111210 10041030A. subglabrum 3102010210 3101111210 10021030A. glabrum nudum 3102010010 3101111210 10021030A. glabrum glabrum 3102010210 3101111210 10021030A. euphorbiae 1102010010 4101111210 12021030A. affine 1102010010 4101111210 12021030A. jubae jubae 1102010010 1101111200 12021030A. jubae tuberculatum 1102010010 1101111200 11021030A. aeoni 1102010011 2101111210 11021030A. wollastoni 1102010011 2101111210 12021030A. bicolor 1102010110 0101111210 12021030A. hesperidum 1102010010 2101111210 10021030A. piscatorium 1102010011 2101111210 10021030


AS on M


Dow

nloaded from


Date post:	15-Aug-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

Syst. Biol. 53(5):711–734, 2004 ISSN: 1063-5157 print ... · Syst. Biol. 53(5):711–734, 2004...

Documents